An electrostatic MEMS switch is a switch operated by an electrostatic charge and manufactured using micro-electro-mechanical systems (MEMS) techniques. A MEMS switch can control electrical, mechanical, or optical signal flow. MEMS switches have typical application to telecommunications, such as DSL switch matrices and cell phones, Automated Testing Equipment (ATE), and other systems that require low cost switches or low-cost, high-density arrays.
As can be appreciated by persons skilled in the art, many types of MEMS switches and related devices can be fabricated by either bulk or surface micromachining techniques or a combination of both types of techniques. Bulk micromachining generally involves sculpting one or more sides of a substrate to form desired three-dimensional structures and devices in the same substrate material. The substrate is composed of a material that is readily available in bulk form, and thus ordinarily is silicon or glass. Wet and/or dry etching techniques are employed in association with etch masks and etch stops to form the microstructures. Etching is typically performed on the backside or frontside of the substrate. The etching technique can generally be either isotropic or anisotropic in nature. Isotropic etching is insensitive to the crystal orientation of the planes of the material being etched (e.g., the etching of silicon using a mixture of hydrofluoric, nitric, and acetic acids (HNA) as the etchant). Anisotropic etchants, such as potassium hydroxide (KOH), tetramethyl ammonium hydroxide (TMAH), and ethylenediamine pyrochatechol (EDP), selectively attack different crystallographic orientations at different rates, and thus can be used to define relatively accurate sidewalls in the etch pits being created. Etch masks and etch stops are used to prevent predetermined regions of the substrate from being etched.
On the other hand, surface micromachining generally involves forming three-dimensional structures by depositing a number of different thin films on the top of a silicon wafer, but without sculpting the wafer itself. The films usually serve as either structural or sacrificial layers. Structural layers are frequently composed of polysilicon, silicon nitride, silicon dioxide, silicon carbide, or aluminum. Sacrificial layers are frequently composed of polysilicon, photoresist material, polymide, metals, or various kinds of oxides, such as PSG (phosphosilicate glass) and LTO (low-temperature oxide). Successive deposition, etching, and patterning procedures are carried out to arrive at the desired microstructure. In a typical surface micromachining process, a silicon substrate is coated with an isolation layer, and a sacrificial layer is deposited on the coated substrate. Windows are opened in the sacrificial layer, and a structural layer is then deposited and etched. The sacrificial layer is then selectively etched to form a free-standing, movable microstructure such as a beam or a cantilever out of the structural layer. The microstructure is ordinarily anchored to the silicon substrate, and can be designed to be movable in response to an input from an appropriate actuating mechanism.
Many current MEMS switch designs employ cantilevered beam/plate, or multiply-supported beam/plate geometry for the switching structure. In the case of cantilevered beams/plates, these MEMS switches include a movable, bimaterial beam comprising a structural layer of dielectric material and a layer of metal. Typically, the dielectric material is fixed at one end with respect to the substrate and provides structural support for the beam. The layer of metal is attached on the underside of the dielectric material and forms a movable electrode and a movable contact. The layer can be part of the anchor or attachment to the substrate. The movable beam is actuated in a direction toward the substrate by the application of a voltage difference across the electrode and another electrode attached to the surface of the substrate. The application of the voltage difference to the two electrodes creates an electrostatic field which pulls the beam towards the substrate. The beam and substrate each have a contact which can be separated by an air gap when no voltage is applied, wherein the switch is in the “open” position. When the voltage difference is applied, the beam is pulled to the substrate and the contacts make an electrical connection, wherein the switch is in the “closed” position.
One of the problems that faces current MEMS switches having a bimaterial beam is curling or other forms of static displacement or deformation of the beam. The static deformation can be caused by a stress mismatch or a stress gradient within the films. At some equilibrium temperature, the mismatch effects can be balanced to achieve a flat bimaterial structure, but this does not correct the temperature effects. The mismatch can be balanced through specific processes (i.e., deposition rates, pressures, methods, etc.), material selection, and geometrical parameters such as thickness. This bimaterial structure of metal and dielectric introduces a large variation in function over temperature, because the metal will typically have a higher thermal expansion rate than the dielectric. Because of the different stales of static stress in the two materials, the switch can be deformed with a high degree of variability. Switch failure can result from deformation of the beam. Switch failure can occur when (1) electrical contact cannot be established between the movable and stationary contacts, (2) electrical contact is established without any actuation, or (3) the operational parameters are driven out of the acceptable specification range due to static deformation or because of the deformation introduced as a function of temperature. A second mode of failure is observed when the movable contact and the stationary contact are prematurely closed, resulting in a “short”. Because of the deformation of the beam, the actuation voltage is increased or decreased depending on whether it is curved away from the substrate or towards the substrate, respectively. Because of this variability, the available voltage may not be adequate to achieve the desired contact force and thus contact resistance.
Some current MEMS switch designs having the bimaterial beam attach the metal layer for the movable electrode to the topside of the dielectric material. The metal layer for the moving contact must still be on the underside of the dielectric material. This design can in some instances serve to provide isolation between the movable electrode and the stationary electrode on the substrate; however, this design requires a higher voltage for actuation because the gap distance between the metal layer and the electrode attached to the surface of the substrate is greater. The effective gap is the sum of the gap between the stationary electrode and the dielectric, and the dielectric thickness. Thus, such a design requires greater power consumption and creates problems with regard to dielectric charging.
A common approach to develop a cross-bar switch array is by a process of forming the cross-bar interconnect structure on a printed wire board (PWB), printed circuit board (PCB), low temperature ceramic composite (LTCC) substrate, or polymer composite substrate and subsequently attaching a switch to the board or substrate. The switch can be attached by a combination of methods such as soldering, wire bonding, bump bonding, flip chip, and other attachment and electric interconnection methods. In this process, the fabrication of the cross-bar interconnect structure is integrated with the MEMS switch process so that they are fabricated on the same substrate with the same process. The advantage of the cross-bar interconnect structure is that an array of input signals can be electrically communicated to a single (or multiple) output of an array of outputs. The array dimensions can be square or rectangular, wherein a square array is an NXN array of N inputs and N outputs and a rectangular array is an NXM array of N inputs and M outputs. The input and output lines of a cross-bar interconnect are electrically connected by the MEMS switch when it is activated in a “closed” position. At each switching node in the cross-bar array, the input/output lines have stationary contact terminals. The integrated MEMS switch functions to establish mechanical and electrical connection between the moving contact and stationary contacts of the input and output lines. The input and output lines communicate through the contact established by the MEMS switch. Alternatively, MEMS switches can be used in a cross-bar array in the usual fashion, if it is attached to the switching nodes of the cross-bar array. This configuration is undesirable because a highly capable fabrication process must be replaced by a less capable assembly process. This will increase the total cost, decrease the quality, complicate the process, and increase the size.
Therefore, it is desirable to provide a fabrication process for producing a structural element (e.g., a beam, plate, and membrane) to improve yield, performance over temperature, actuation, and quality of MEMS switches, it is also desirable to provide a fabrication process for producing a structural element resistant to deformation caused by parasitic or “self actuation. It is also desirable to provide a fabrication process for producing a structural element that is robust to process variations, film stresses, and their gradients. Furthermore, it is desirable to provide a method for forming a cross-bar interconnect and MEMS switch in the same fabrication process. It is desirable to provide a method that is flexible so that the MEMS switch can be formed and integrated with cross-bar interconnects that are fabricated in another fashion.